zbMATH — the first resource for mathematics

High-order multi-variable intuitionistic fuzzy time series forecasting model. (Chinese. English summary) Zbl 1363.62121
Summary: In order to break the limitation of fuzzy set theory and objectively describe the uncertain data, a high-order multi-variable intuitionistic fuzzy time series forecasting model is proposed. A fuzzy clustering algorithm is adopted to partition the universe of discourse, and a more objective method is used to establish the membership and non-membership functions of intuitionistic fuzzy sets. According to the principle of multidimensional intuitionistic fuzzy modus ponens reasoning, a heuristic similarity-based reasoning technique is proposed as the forecasting rule of the high-order multi-variable forecasting model, and a corresponding defuzzification method is presented. Contrast experiments on the daily mean temperature of Beijing were carried out. Experimental results show that the root mean square error (0.86) and the average forecasting error ($$2.57\%$$) of the proposed model both obviously decreased. Therefore, the forecasting performance of the model is better than that of the fuzzy time series forecasting models and the normal intuitionistic fuzzy time series forecasting models.
MSC:
 62M86 Inference from stochastic processes and fuzziness 62M10 Time series, auto-correlation, regression, etc. in statistics (GARCH) 62M20 Inference from stochastic processes and prediction
Full Text: